What defines the upper bound for FX resolution/pixel density?

Started Aug 26, 2013 | Discussions thread
The_Suede Contributing Member • Posts: 652
Re: 100/200/500/700/1000 MPixels

Very good and concise post BTW, thanks.

With all this talk about diffraction (that may be educational, but IS starting to get a bit boring by now...) I think it's time for quite a lot of people to get out of their little cage and look at pure results in stead. To do this you sometimes have to take a few steps back from the "problem" and approach it from another angle.

The point to which I intend to get to can be summarized like this:

-"When does my sensor stop limiting my images?"

Today, we get very good both read noise and efficiency from rather small pixels - what still needs work in very-small-pixels is angle-efficiency so that we don't lose light at very large apertures and with semi-symmetrical wide angles (in mirrorless cameras).

IMO no Bayer based camera produces pixel-perfect images straight from full raw resolution. You need to downsample to about 70% scale before each pixel in the image has good or even "very good" correlation to the scene you actually shot.

So, even if I have a lens that on average use, in my average shooting situation would limit my "accurate output information amount" to say 20MP due to aberrations and diffraction, I'd need a 40MP camera to actually utilize that lens fully. Those 20MP's of accurate information present in the image projected by the lens on to the sensor needs 40MP's of Bayer capture resolution to be accurately captured and converted into an image.

I agree on your "somewhere around 100MP" as being the sweet spot for a consumer-oriented 36x24mm sensor camera. This comes from having worked with K-8 MTF data up to 120lp/mm for quite a few years, and having some background in electroptics - I don't see any really relevant improvements to the end image from going further than that (for general use, of course... Always include a statement intention limiter... )

What I do sorely miss is a very competent interpolation engine in camera internals, so that each user can set his/hers raw-resolution. But one of the positive things about pushing the resolutions higher is that then we don't have to rely on advanced guessing software algorithms to build the image. When the image resolution approaches the optical limits the interpolation engine is taken out of the image quality equation. At that point you don't have to ASK the engine for additional (and uncertain!) detail - which is the main problem with raw-converters today.

A raw converter today tries to guess and fill in detail that "probably" could have been there. Being able to relax how aggressively the algorithms tries to create image detail also lowers the amount of digital interpolation artifacts by quite a lot.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow